Effect of Mammography on Breast Cancer Mortality

Clinical Scenario

A 54-year-old woman is new to your practice and has never been screened for breast cancer. She is concerned about false-positive results, and wonders if mammography does more harm than good.

Clinical Question

Do the potential benefits of mammography outweigh the potential harms?

Evidence-Based Answer

Although mammography can lead to false-positive results, emotional stress, and unnecessary biopsies, it appears to reduce deaths from breast cancer by about 15 percent.^[1] (Strength of Recommendation: A, based on consistent, good-quality patient-oriented evidence.)

Practice Pointers

Breast cancer is the second most commonly diagnosed cancer (after skin cancer) in U.S. women.^[2] Most primary care physicians recommend screening mammography in women every one to two years starting at 40 to 50 years of age. Widespread use of mammography and improvements in breast cancer treatment are thought to be responsible for the drop in breast cancer mortality since about 1990.^[3] In 2009, the U.S. Preventive Services Task Force (USPSTF) reevaluated its recommendation on mammography that was released in 2002.^[4] After reviewing new and existing literature,^[5] the USPSTF now recommends against routine screening mammography in women 40 to 49 years of age, and encourages physicians to make the decision to screen these women on a patient-by-patient basis.^[6] The review accompanying the USPSTF recommendation found a mortality benefit in all women, but an increased number of false-positive results in women 40 to 49 years of age.^[7] Women in this age group also have a lower incidence of breast cancer. This Cochrane review examines the risks and benefits of mammography for all age groups.^[1]

The authors examined eight clinical trials comparing the effect of mammography with no mammography on breast cancer mortality and morbidity. One trial was excluded because of bias. The excluded trial reported different socioeconomic profiles of the control and intervention groups, and had differing exclusion criteria for each group.^[8] The remaining seven trials (n = 600,000) were divided into two groups: three trials with adequate randomization that showed a 10 percent reduction in breast cancer mortality (not statistically significant), and four trials with suboptimal randomization that showed a 25 percent reduction in breast cancer mortality (statistically significant).

The problems with randomization included insufficient data to determine who was excluded from the intervention and control groups and why, and in some cases, insufficient information about how practices or participants were randomized. It was not clear if the Cochrane authors knew whether the trials had employed suboptimal randomization practices in every case, or whether there was insufficient information about every step of the process. If the latter is true, then the most conservative approach would be to exclude those trials from the analyses, although it is possible that some or all of the randomization practices in those trials were acceptable. The overall risk reduction in breast cancer mortality for all seven trials was statistically significant (19 percent), but the reduction in all-cause mortality was not statistically significant